Kenneth M. Merz

Computational approaches applied to the drug discovery pipeline are invaluable because of their overall speed or time to solution, sustainability, accessibility combined with ever improving accuracy. Furthermore, the high-throughput virtual screening supports practical sampling of the vast chemical space for target-relevant bioactive compounds that can be experimentally validated further downstream in the drug development process. The current paradigm being practiced in computer-aided drug screening strategies centers around structure-based and ligand-based approaches supported by a range of computational tools including de novo design, molecular dynamics-based approaches, and machine learning (ML). Overall, the latter approach has shown great promise generating models of high computational speed coupled with good accuracy. In this work, we developed QRxVision, a domain-specific multimodal ML model explicitly optimized for kinase target assignment using similarity-based profile matching and retrieval. QRxVision utilizes image vision from a pretrained VGG16 (Visual Geometry Group–16) model for deep spatial feature extraction and cosine similarity calculation. By encoding drug multimodality-associated parameters (MAP) into standardized QR code images, QRxVision can evaluate the similarity between test and reference compounds. Drug targets can then be assigned via transitive profile matching. Validation using kinase inhibitors shows that QRxVision is 71% successful at assigning viable small molecules within the Top-5 predictions (p-value < 0.05). Overall, we believe that with the appropriate algorithmic modifications and domain retargeting, the QRxVision framework can be applied to other target assignment tasks.

In this work, we extended the book-ending framework with a hybrid quantum-classical workflow that incorporates configuration interaction (CI) calculations into alchemical free energy (AFE) predictions. In the book-ending approach, the Multistate Bennett Acceptance Ratio (MBAR) is applied along a coupling parameter λ to interpolate the system from molecular mechanics (MM) (λ = 0) to a quantum mechanics (QM) (λ = 1) description, and the resulting correction is added to the classically computed AFE. Building on the standard book-ending workflow, we developed an interface that introduces the CI contribution through two backends: (I) a classical PySCF-based backend; (II) a quantum-centric sample-based quantum diagonalization (SQD) method and its extended version (ext-SQD). This latter approach combines real quantum processing units (QPUs) with classical postprocessing to obtain CI energies and gradients. To validate the proposed infrastructure, we computed the book-ending corrections for the hydration free energies (HFEs) of three small organic molecules: ammonia, methane, and water. These benchmarks demonstrate that the CI-level electronic structure calculations, particularly those performed on a quantum hardware, can be naturally incorporated into AFE workflows. Specifically, the CI-corrected HFEs are in reasonable agreement with experimental values, supporting the feasibility of QPU-accelerated free energy predictions. As quantum devices continue to improve in scale and fidelity, they might offer a practical and scalable route to CI-quality electronic-structure data for systems that are challenging for classical approaches. Integrating these CI energies directly into QM/MM simulations could improve the accuracy of free energy methods for systems where electronic correlation plays a significant role, with potential relevance to large biomolecular systems, enhancing our ability to model molecular recognition, enzyme catalysis, and drug–receptor interactions.

We present the implementation of a fragment-based, quantum-centric supercomputing workflow for computing molecular electronic structure using quantum hardware. The workflow is applied to predict the relative energies of two conformers of the 303-atom Trp-cage miniprotein. The methodology employs wave function–based embedding (EWF) as the underlying fragmentation framework, in which all atoms in the system are explicitly included in the configuration interaction (CI) simulations. We employ sample-based quantum diagonalization (SQD) solver for challenging fragments and full configuration interaction (FCI) solver for trivial fragments. The EWF-(FCI,SQD) results are compared against EWF-MP2 and EWF-CCSD benchmarks. The impact of fragmentation on the predicted relative energies of the Trp-cage conformers is further evaluated by comparison with unfragmented RI-MP2 and DLPNO-CCSD calculations. The results demonstrate that large-scale electronic configuration interaction (CI) simulations of protein systems containing hundreds or even thousands of atoms can be realized through the combined use of quantum and classical computing resources.

Nramps are a family of proton-coupled divalent metal ion transporters that play critical roles in maintaining homeostasis of essential metals. In humans, Nramp2 (DMT1) mediates iron uptake to support both cellular and systemic iron homeostasis, whereas Nramp1 exports metals from phagosomes, contributing to antimicrobial defense. A survey of experimentally solved Nramp structures reveals that an N-terminal helix (αA) immediately preceding TM1 is folded only in the outward-facing or outward-facing occluded states, raising the question of its role in Nramp transport. Here, we combined mutagenesis, transport assays, and molecular dynamics simulations to investigate αA. Our results show that deletion or substitution of conserved residues in αA markedly reduces Fe ²⁺ transport. Simulations indicate that αA behaves as an amphipathic helix in the outward-facing state and preferentially stabilizes this conformation over the inward-facing state, thus acting as a key player in controlling conformational equilibrium. This study provides new insights into structural dynamics of Nramps and potentially sheds light on other LeuT-fold transporters.

General force fields such as General Amber Force Field (GAFF) have been designed for broad applicability and are widely used in protein-ligand binding simulations in structure-based drug discovery. However, the force field parameters are not always transferable across ligand molecules, and custom reparameterization is sometimes necessary for accurate binding free energy simulations. This is especially true for torsion parameters, which are highly dependent on stereoelectronic and steric effects. Here, we report a novel, flexible, and user-friendly computational tool called the Automated Force Field Developer and Optimizer (AFFDO) platform that allows generating accurate, tailored GAFF2 torsion parameters for drug-like molecules. For a given ligand, AFFDO selects the most important torsions, carries out GPU-accelerated density functional theory calculations to collect reference data and fits torsion terms using a fast gradient-based optimizer that leverages automated differentiation. We benchmark AFFDO by parametrizing a series of drug-like molecules and carrying out protein-ligand relative binding free energy (RBFE) simulations. The results show that AFFDO can significantly improve GAFF2 torsion parameters against QM reference data, which in some cases translates into better agreement with experimental RBFE values within a reasonable computational time.

We describe the implementation details of highly efficient ab initio electrostatic potential (ESP) calculations on graphics processing units (GPUs) and introduce a novel scheme for partial charges that are robust against molecular orientation. Performance analyses are discussed, and we highlight that in our new implementation, a single data center GPU can outperform 128 corresponding data center CPU cores in time to solution. This implementation in the open-source Quantum Interaction Computational Kernel code (QUICK) enables ESP computations on highly dense grids that surpass what is reported in the literature, on the order of N_grid points ∼ 20000 points/atom. We demonstrate that, in this dense-grid limit, ESP charges become independent of molecular orientation. We denote such ESP charges as being robust against molecular orientation and validate this desirable attribute against standard charge schemes. Our proposed charge scheme, called reweighted RESP (rwRESP), is designed to significantly overcome the sensitivity to N_grid points that limits the reliability of canonical RESP charges. By effectively amending this N_grid points-sensitivity, we demonstrate that rwRESP charges also achieve robustness against molecular orientation. Ultradense-grid ESP computations and rwRESP fits can be readily performed via the seamless integration of QUICK with AmberTools, enabling highly efficient and reliable parametrization of the general AMBER force field (GAFF) for nonstandard residues. In this spirit, we believe that our fully fledged GPU protocol for obtaining robust molecular charges will facilitate a wide range of applications, such as high-throughput parametrization of molecular interaction potentials, while also serving as a foundational step toward GPU-accelerated on-the-fly polarizable QM/MM simulations with QUICK.